An AI assistant that drafts notes from dictation
After the visit you speak, and the AI assistant turns your words into a structured note draft that fits your specialty, flagging missing fields and contradictions. You edit, approve and sign the draft; nothing becomes part of the clinical record without physician approval.
The assistant produces drafts, warnings and summaries
Every output is a draft or a suggestion; none of it is published directly as a record, prescription or patient message.
- Dictation, note draft and summary
- Speech is transcribed and a structured note draft (for example SOAP) is created by specialty and visit type. A summary of a long patient file or consultation is produced and also kept as a draft.
- Code and template suggestions, warnings
- A diagnosis or procedure code or template suggestion that fits the note is offered. Contradictions, missing fields and follow-up work, such as a check-up appointment, are flagged as warnings. A suggestion does not reach the record until accepted.
- Operational insight
- For operational questions such as no-show trends, capacity and appointment load, it produces a summary and a suggestion that gives management an overview.
Dictation
The patient came in for a check-up. Complaints have decreased. Blood pressure readings are within normal limits. Keep the same dose. Follow-up in three weeks.
Structured draft
- S
- History
- Complaints decreased
- O
- Findings
- Blood pressure within normal limits
- A
- Assessment
- To be completed by the physician
- P
- Plan
- Same dose, follow-up in three weeks
Missing field: allergy history
- Edit
- Approve and sign
- Reject
The dictation and draft on screen are examples prepared for illustration.
From dictation to signature, the draft is finalized by the physician
At every step of the flow the decision rests with the physician. The AI prepares a draft and the physician edits it; a rejected draft never enters the clinical record. If the AI is off, the same flow continues with the note the physician writes. After approval, the source text, the model output, the physician's changes and the approval trail are stored together.
Source and trail stored
The source text, model output, the physician's changes and the approval trail are kept together.
- The process is complete here.
All steps
- 01 Dictation (Physician) After the consultation the physician speaks; the audio is transcribed against a specialty-appropriate template.
- 02 Structured draft (Assistant) The speech becomes a structured note draft for the specialty and visit type. The draft is marked as a "suggestion".
- 03 Missing-field and contradiction warning (Assistant) Empty required fields, contradictory statements and follow-up jobs are flagged as warnings.
- 04 Physician edits (Physician) The physician corrects the draft, resolves the warnings or declines the suggestion.
- 05 Approval and signature (Physician) The physician approves and signs; the record is finalized only now.
- 06 Source and trail stored (Record) The source text, model output, the physician's changes and the approval trail are kept together.
Alternative path from 04 Physician edits; the process ends on this path.
- 04a Rejected: not finalized (Physician) The physician does not accept the draft; it never enters the clinical record and the physician writes the note themselves.
Alternative path from 01 Dictation; it returns to the main flow at 04 Physician edits.
- 01a AI off: continue manually (Record) If the assistant is off, the physician writes the note directly; appointments, signatures and other flows are unaffected.
06 Source and trail stored
The time spent on writing notes comes back
The visit ends and the tiring part begins: writing the note. A complete, timely note in a structure that varies by specialty can take a physician's evening. The assistant lightens that load: it turns your speech into a structured draft, reminds you of an empty field and flags follow-up work. Reading and correcting a draft is shorter than writing from scratch, and the last word is still yours.
It is designed for physicians who carry a writing load after busy clinics, specialists who need to review long patient files quickly, and clinics that want structured specialty templates filled in completely. When it is switched off, no workflow is missing; the physician writes the note as usual.
An approved record carries not only the final text but also which draft it came from and what the physician changed. When someone asks how a record came about, the answer comes from the trail, not from memory.
Clinical recordsAI is an assistant; authority, decision and responsibility stay with the physician
Physician approval is not a setting but the way the system is built: the AI cannot write to the database, cannot run a command on its own, and cannot publish an output as a record, prescription, order or patient message.
- Physician signature
- A clinical record is not finalized without physician approval; a signature is solely the physician's action.
- The decision is the physician's
- Diagnosis, treatment and prescription are determined by the physician; the assistant produces note drafts, summaries, warnings and suggestions.
- Trail
- The source text, model output, physician's changes and approval trail are kept, so how a record came about can be followed afterwards.
- Not used for training
- Health data is not used to train models.
- Controlled processing
- Patient and clinical content is routed to be processed in an environment under HekimBis's control, not sent to a general-purpose external service.
- A switch per scenario
- Each use case is switched on or off by plan, organization and purpose. In a switched-off scenario no data is processed, and appointments, visits, signatures, collections and other critical flows run independently of AI.
Frequently asked questions
Does the AI make diagnoses?
The assistant does not diagnose, determine treatment or prescriptions, or read diagnoses from images. It only produces note drafts, summaries, warnings and suggestions; the decision and the signature belong to the physician.
Can anything enter the record without physician approval?
AI output is stored as a draft. For it to become a record, the physician must edit and approve it and sign it with a re-authorized action. A rejected draft never enters the clinical record.
Is my patients' data used to train the AI model?
No. Health data is not used to train models.
Does my data go to an outside service?
Patient and clinical content is routed to be processed in an environment under HekimBis's control, not sent to a general-purpose external service. Each use case can be switched on or off separately by organization, plan and purpose; in a switched-off scenario no data is processed.
Does the system work while AI is switched off?
Yes. Appointments, visits, signatures, collections and every other critical workflow run independently of AI. When it is off, only assistive features such as draft notes from dictation and summaries are unavailable.
Which languages does dictation work in?
Supported languages and specialty templates are confirmed in the demo together with your clinic's specialty.
Can I see afterwards how an AI output came about?
Yes. The source text, model output, physician's changes and approval trail are stored together, so how a record came about can be followed afterwards.
Can the AI send messages to patients?
AI output is not published to patients as a message. An output is a draft or a suggestion; publishing needs a separate user action and re-authorization.
Is AI usage charged?
AI model and transcription usage is one of the metered items; scope and unit price are set by the plan and add-on model. Current information is on the pricing page.
Related pages
Features
Specialties
Software guides